Keynote: Sovereign AI at Population Scale...
summary
Sarvam builds a full India-first stack—Sarvam 105B and 30B LLMs, Bulbul TTS, Saras STT, Samvada agents and Prava token factory—running on a single Kubernetes GPU fleet with KEDA autoscaling, suffix/prefix model caching and on‑prem data sovereignty. The platform serves 10 million API calls daily, achieves sub-second latency for voice, digitizes massive document volumes and enabled 17 million farmers in 24 days using CNCF components like Linkerd, Strimzi, Argo CD, Prometheus, Thanos, CloudNativePG.
AI-written summary · about
related talks
24mKeynote: Building India’s Sovereign AI Stack w/ Local Talent & Open Global Collab
01h 07mKeynote + Conversation
29mOpen Source Sovereign AI: Building Enterprise GenAI Platforms for Control & Comp...
34mKeynote: Building for AI at Scale: Infra, Integrity, & Innovation
35mSovereign AI in Action: Advancing India’s Digital Future With Ope...
41mKeynote
09mSponsored Keynote: Production AI Runs on Open Sou...
09mKeynote: Kubernetes for GenAI: The New Operating System of the AI-First World
